what a custom three.js site actually costs

You searched this because you got a quote that scared you, or you're about to ask for one and you want to know if you're about to get robbed. Fair. The whole market is built to keep you from knowing the answer, so here it is, in real numbers, from one person who builds these for a living and has no boss telling them to round up.

A custom Three.js site — meaning a site with real 3D in the browser, not a stock template with a spinning logo — runs somewhere between six thousand and well over a hundred thousand dollars. That range is useless, I know. The whole point of this page is to cut it into pieces that mean something, and to tell you which piece you're actually buying, because most of the people quoting you are happy to let you think you need the expensive piece.

A few ground rules before the numbers, because the numbers lie without them. There's no "starting from." A real quote is a range with a top, and if someone gives you a floor and no ceiling, they're planning to find the ceiling on your invoice. And these are build prices — the cost to make the thing once, not the much smaller cost to keep it running after.

the three real tiers

A single well-built long-page with one 3D moment. Roughly $6k–15k.

One page. Scrolls. Has one genuine piece of 3D that makes you go oh — a product that rotates as you scroll, a hero scene that reacts to the cursor, one thing that's actually rendered in WebGL and not a video pretending to be. Custom design, not a theme. This is most of what people actually need and most of what people get talked out of.

At the bottom of that range you're getting one good moment and clean everything-else. At the top you're getting a moment that's been fussed over — real lighting, a model that doesn't melt a phone, motion that feels intentional instead of stock. The gap between $6k and $15k is almost entirely how much obsessing happens over that one moment. The obsessing is the product. It's also the first thing a cheap shop skips.

A real multi-page site with a configurator or a store. Roughly $18k–45k.

Now it's a system. Several pages, a content setup so you can update it without calling anyone, and one piece of heavy interactive 3D that does a job — a product configurator where people build their own version and watch the price move, a store where the 3D is tied to what they're buying, an item viewer wired into checkout. The 3D isn't decoration here. It's the thing that makes people open their wallet.

The money in this tier goes somewhere specific, and it's worth knowing where. It's not the pages. Pages are cheap. It's the configurator logic — every option that has to know about every other option, every combination that has to render right and price right and not break on the one weird selection nobody tested. That's the hard part, that's where the hours go, and that's the part that looks like five minutes of work to anyone who hasn't built one.

A from-scratch 3D centerpiece — the thing is the showpiece. Roughly $45k–120k, and up.

This is when the 3D is the site. A navigable world, a custom WebGL experience, something nobody else has because nobody else built it. Custom shaders, a real asset pipeline, performance work so it runs on a four-year-old laptop, the works. This is a real chunk of someone's year.

Honest warning, because this is where the most money gets set on fire: most people who want this tier need the first tier. The world you can fly through is incredible and it's also five weeks nobody clicks past the second region. If you're a small brand, you almost certainly do not need this, and the shop that nudges you toward it is quoting their dream project on your budget. There are exactly the right reasons to build it — a game, an art piece, a launch where the experience is the product — and a lot of wrong ones, and "it would look amazing" is a wrong one all by itself.

where the money actually goes

Strip away the tier and three things drive every price.

• The 3D and WebGL hours. This is the expensive labor and there's no faking it. Making a model run smoothly in a browser, on a phone, without looking cheap, is genuinely hard and genuinely slow. If a quote is low and promises a lot of 3D, the 3D is going to be bad. There's no third option.
• The custom design. A site that looks like nobody else's takes someone with taste making a hundred decisions you'll never see. This is real money and it's the money most worth spending, because it's the part you literally cannot get from a template.
• The back-and-forth. Revisions, the "can we try it blue," the founder who changes their mind in week three. This is real and it's fine, but it's hours, and a low quote usually means it isn't budgeted — which means it comes back as a change order later, when you've already paid and have nowhere to push back.

what you're overpaying for

Now the part a vendor won't put in writing.

• Agency overhead on junior hands. You're quoted the senior rate, the work gets done by whoever's free, and a project manager forwards emails between you and the person actually building. You're paying three salaries for one person's output. A studio of one or two, where the person you talk to is the person typing, costs less and is usually better, because nothing gets lost being relayed.
• "Enterprise" tooling you'll never fill. A headless setup built for a thousand editors when you have one. A CMS license priced for a corporation. Infrastructure for traffic you don't have. A small brand needs a small, fast stack. Anything sold as "scalable" before you have anything to scale is sold to you, not for you.
• The pitch. Some of what you pay funds the next client's sales deck. The polished proposal, the discovery workshop, the slide that says "discovery workshop." A small shop skips the theater because the theater is the overhead, and the savings are real.

so what should you actually do

Start one tier below where you think you are. Most people who think they need the world need the great long-page. Build that, ship it, see what it does, then spend on more 3D once you know people want it — not before. The cheapest version of this whole thing is the one where you don't build the impressive part first.

I'm building a small thing to let you ballpark your own number without talking to anyone. It isn't ready, so I won't pretend it is. Until then, the honest move is to take your real scope to one person who builds these, not a sales team, and ask them to talk you down a tier. The good ones will. That's how you find a good one.

And if you'd rather just describe what you're making and get a straight number — hello@nightshiftglow.studio. No deck. No discovery call. Just tell us what it is.